Austin
Tam
Projects
Projects
Conducted sales data analysis and customer
profiling to better inform marketing strategies
Generated insights through customer analysis to
inform expansion strategy
Optimized asset allocation by analyzing seasonality and other
determining factors of demand
Updated the company’s global market
understanding based on historic data
Provided pricing insights to inform user and provider expectations
and to guide relevant product decisions
Projects
Provided pricing insights to inform user and provider expectations and
to guide relevant product decisions
Conducted sales data analysis and customer
profiling to better inform marketing strategies
Generated insights through customer analysis to
inform expansion strategy
Optimized asset allocation by analyzing seasonality and other
determining factors of demand
Updated the company’s global market
understanding based on historic data
Airbnb Vacation Analysis
Analyze Airbnb data
from nine different
European cities to
discover insights for
both providers and
users on Airbnb.
Namely, help
providers and users
which features drives
prices, and
understand
difference between
different cities
•Data wrangling,
visualizations,
Clustering,
geospatial analysis,
Time series analysis
and machine
learning in Python
•Storyboard creation
in Tableau
•Markup and
notebook
management in
Jupyter
The data that was
used for this
analysis was
published as open
source from
Eurostat and from
an academic paper
which can be found
here and here.
Our analysis using the
Eurostat data is limited
by granularity,
specifically that the
data is at a country level
and not a city level. In
terms of the other
dataset, it has
limitations due to its like
of a temporal variable
and its limited number
of variables.
Exploratory Data Analysis
After cleaning and merging datasets, the dataset was explored to find any patterns or
preliminary insights
Clustering & Geospatial Analysis
I began to dive deeper
into the data using
clustering and geospatial
analysis techniques to
look for patterns within
the data, and to look at
the geographic aspect of
the data.
Regressions & Time-Series Analysis
To further the analysis, we modeled the data using regressions and conducted a time
series analysis
Insights for the
The full project can be found on my GitHub here and the accompanying Tableau storyboard can be found here.
Airbnb Analysis
This analysis of European Airbnb data resulted in insights for both providers and
guests that aim to better their respective user experience. Such insights mainly touch
upon an Airbnb’s features and how they impact the price of the Airbnb.
Projects
Provided pricing insights to inform user and provider expectations and
to guide relevant product decisions
Conducted sales data analysis and customer
profiling to better inform marketing strategies
Generated insights through customer analysis to
inform expansion strategy
Optimized asset allocation by analyzing seasonality and other
determining factors of demand
Updated the company’s global market
understanding based on historic data
InstaCart Basket Analysis
Analyze InstaCart’s
user data by
conducting an initial
data and exploratory
analysis and applying
machine learning
models in order to
derive insights and
suggestions for
InstaCart’s upcoming
marketing campaign, as
well as to increase
overall sales for the
company.
•Data wrangling,
creating
visualizations, and
machine learning in
Python using
libraries such as
pandas, seaborn,
and sklearn
•Report creation in
Excel
•Markup and
notebook
management in
Jupyter
The data that was used
for this analysis was
published as open
source from InstaCart
and can be found here.
The customer and
demographic data was
fabricated for the
purpose of this analysis
and can be downloaded
here.
Our analysis is limited in
part by the fabrication
of customer data. In
addition, the dataset is
only from 2017 and so it
hinders any kind of
temporal analysis as
well as any efforts to
mitigate any temporal
bias. Lastly, there is
limited customer data,
thus only so many
customer profiles were
made.
Data Cleaning, Wrangling & Merging
Multiple datasets were cleaned, wrangled and then merged to create the final dataset
Exploratory Data Analysis
Creating various visualizations helped us to see trends in customer
spending habits
Machine Learning
In addition, a Market Basket
Analysis was done in order to find
the most popular item triplets
An XGBoost classifier was
created to determine which user
characteristics lead the user to be
a high or low spender.
Insights from InstaCart Basket Analysis
This analysis of InstaCart’s data revealed many different aspects of customer spending habits
including: most popular times of the day and week in terms of ordering, most popular products and
departments, and customer brand loyalty.
The full project can be found on my GitHub here.
Projects
Provided pricing insights to inform user and provider expectations and
to guide relevant product decisions
Conducted sales data analysis and customer
profiling to better inform marketing strategies
Generated insights through customer analysis to
inform expansion strategy
Optimized asset allocation by analyzing seasonality and other
determining factors of demand
Updated the company’s global market
understanding based on historic data
Media Retail Expansion
This dataset is
provided by
PostgreSQL for
usage in tutorials. It
contains data about
film inventory,
customers,
payments, and
associated details.
The dataset can be
accessed here.
Analyze RockBuster
Stealth LLC’s company
data in order to inform
the company’s
expansion decisions.
Namely to inform their
decision to start a new
online video streaming
platform
•Data visualization
using Tableau
•Analyzing and
merging data using
PostGRE SQL in Pg
Admin 4
•Creation of data
dictionary in
dbdocs.io using
dbml
•Presentation created
in powerpoint
The dataset is provided
by PostgreSQL for
usage in tutorials. It
contains data about film
inventory, customers,
payments, and
associated details.
The data can be found
here
The data is limited by its
manufactured nature,
as it does not contain
any realistic data.
Therefore, while
analysis was possible,
the actual insights
garnered from such
analysis were not very
helpful, and at times did
not make sense.
An entity
relationship
diagram (left)
was created
so that the
tables’
contents and
their
relationships
to one
another could
be seen at a
glance.
ERD & Data Dictionary
Using SQL,
answers to
questions such
as ‘how many
regular
customers and
top customers
are there in each
country?’ were
found.
Afterwards, using
the exported csv
table,
visualizations
such as the one
on the left were
created
SQL & Visualizations
Insights for the
The full project can be found on my GitHub here.
This analysis for
Rockbuster Stealth’s
online video streaming
service resulted in a
myriad of
recommendations,
including: most popular
content to include on
the platform,
geographical locations
to prioritize, most
popular
actor/actresses, and a
list of most valued
customers
Media Retail Expansion
Projects
Provided pricing insights to inform user and provider expectations and
to guide relevant product decisions
Conducted sales data analysis and customer
profiling to better inform marketing strategies
Generated insights through customer analysis to
inform expansion strategy
Optimized asset allocation by analyzing seasonality and other
determining factors of demand
Updated the company’s global market
understanding based on historic data
Influenza Analysis
Identify geographic and
seasonal trends for
influenza in order to
advise a medical
staffing agency on their
staffing decisions for
this upcoming influenza
season. Namely to
advice where and when
to staff medical
personal across the US.
•Data profiling and
cleaning
•Data research and
project design
•Statistical
hypothesis testing
•Geogrpahic
visualizations and
time-series
forecasting
•Interactive
visualizations and
storytelling in
Tableau
The data used comes
from the US Census
Bureau and the CDC.
From the former comes
the Census population
data. From the latter
comes the flu death
reporting, flu lab tests,
flu-like illnesses clinical
visits, and survey of flu
shot rates
While the data has high
quality and integrity
there are a couple
features that constrain
our analysis. One of the
main ones is
geographical
specificity. Another is
that some of the data
sources are surveys
and therefore may
underrepresent certain
groups of people
Hypothesis Testing
Hypotheses were formed
through correlation testing and
exploratory visualizations.
Afterwards, such hypotheses
were tested using t-tests.
Hypothesis Testing
The two variables that were
used to determine where to
send extra staff were poverty
rates and percentage of
population over the age of 65.
Determining Factors
Influenza Analysis
Through this analysis both temporal and geographical
recommendations were generated. The chart above reveals
when staff should be sent. And the graph to the right ranks
the states by order of need of staffing.
The full storyboard can be found on my Tableau here.
Insights from
Projects
Provided pricing insights to inform user and provider expectations and
to guide relevant product decisions
Conducted sales data analysis and customer
profiling to better inform marketing strategies
Generated insights through customer analysis to
inform expansion strategy
Optimized asset allocation by analyzing seasonality and other
determining factors of demand
Updated the company’s global market
understanding based on historic data
The data used in this
project was made
publicly available by
VGChartz. It covers
historical retail sales of
videogames for games
that sold more than
100,000 copies up to
the year 2016.
Company Sales Analysis
Using the company’s
data, test whether or
not the company’s
current geographical
sales assumptions are
correct. If they are not,
use the resulting
insights to inform the
company’s marketing
strategy for the
upcoming year.
•Data profiling and
cleaning
•Pivot tables in Excel
•Descriptive analysis
•Visualizations in
Excel
•Reporting in
Powerpoint
The main constraint in
this dataset is that it
does not contain any
data about any online
videogame sales.
Therefore, the following
analysis only pertains to
physical video game
sales, which may
account for some of the
trends uncovered in the
insights.
Preliminary Insights
Preliminary analysis revealed that the company’s assumptions about
geographical sales were incorrect. Sales had been varying widely across
regions and declining in some regions as well.
Alongside geographical recommendations, other insights that were
developed during the course of the analysis generated console/device
and gaming genre recommendations as well.
The full presentation can be found here.
Insights from Company Sales Analysis
Thank you
Austin Tam